John Yang(@jyangballin):Multi-agent runs on ProgramBench in Opus 5.5 system card Multi-agent systems get to the same outcomes as a single-agent, but quicker. Worth noting that just 166/200 tasks were used. I still strongly encourage using all 200! The long tail of hard tasks is really difficult.

John Yang

@jyangballin

CS PhD @Stanford. Created @SWEbench (multi-lingual/modal); SWE-agent; SWE-smith; InterCode; CodeClash; ProgramBench

Joined June 2012

1.1K Following 7.3K Followers

John Yang@jyangballin

2026.09.22 19:33

Multi-agent runs on ProgramBench in Opus 5.5 system card Multi-agent systems get to the same outcomes as a single-agent, but quicker. Worth noting that just 166/200 tasks were used. I still strongly encourage using all 200! The long tail of hard tasks is really difficult.